MAHMOUD HAMED ISMAEL

visualization

bivariate

lets insights some category accorging to country destination

use Z-score to remove outliers

summary

Merge train and test datasets 1- in GENDER Unkown value (129480) 47% female(77524) 28% male(68209) 24% other (34) 0.1% 2- top 5 first device type Mac Desktop 106328 Windows Desktop 86948 iPhone 39814 iPad 18036 Other/Unknown 11167 3-signup app Web 219918 iOS 34593 Android 10519 Moweb 10517 4-signup method basic 198222 facebook 74864 google 2438 5- top 5 affiliate provider direct 181270 google 65956 other 13036 facebook 3996 bing 3719 6- top 5 signup flow page 0 206092 25 29834 12 11244 3 8822 2 6881 7- top 5 country destination NDF 124543 (No Destination Found ) 45% US 62376 (22.6%) other 10094 (3.7%) FR 5023 (1.8%) IT 2835 (1%) 8- account created years 2014 (50.26%) 138562 most month july 2013 (30.11%) 82960 most month september 2012 (14.32%) 39462 2011 (4.27%) 11775 2010 (1.01%) 2788 10 -More conclusions in bivariate analysis that graphs shows